Skip to content

PPSA26578: sceLibcBacktraceGetBufferSize not implemented, abort after boot | Windows, i7-13700K, RTX 5060 #8308

Closed

Description

@MurphylawdenINC

Description

After the libc imports reported in #7624 were resolved, PPSA26578 now starts and then aborts. The guest calls sceLibcBacktraceGetBufferSize, which is not implemented, so libc throws and the process terminates.

Error

libc: unhandled exception 'St13runtime_error' thrown from 00007ff89a2b113f
terminate called after throwing an instance of 'std::runtime_error'
what(): sceLibcBacktraceGetBufferSize_nid_postfix not implemented
FATAL: abort() (SIGABRT)

The call comes from the guest (Il2cppUserAssemblies.prx.guest.prx appears in the stack candidates).

Environment

  • AnyPS5 version / commit:
  • OS: Windows 11
  • CPU: Intel Core i7-13700K
  • GPU: NVIDIA GeForce RTX 5060 (driver )
  • RAM: 32GB DDR5
  • Extraction firmware: 13.42 (libs\ extracted from my own console)
  • Game: PPSA26578 (<base / patched, version>), Unity/IL2CPP

Notes

Log

anyps5.log

Activity

  1. MarcZX10 commented on Oct 20, 2026

    @MarcZX10
    Contributor

    An open PR already implements sceLibcBacktraceSelf and sceLibcBacktraceGetBufferSize on Windows: #8137. Its hosted checks pass, but the Windows runtime job is skipped. I left a focused test-coverage suggestion on the PR to verify that the first frame is exactly the caller, not just somewhere in the same module. This issue appears covered by #8137, so there is no need for a duplicate implementation.

  2. JJenkx commented on Oct 22, 2026

    @JJenkx
    Contributor

    #8137, which implements \sceLibcBacktraceSelf\ and \sceLibcBacktraceGetBufferSize\ on Windows, has been merged, and current \main\ (6ef5a1c) has both in \core/libs/prx/libc/src/Backtrace.cpp. A rebuild from current \main\ should get past this abort. I haven't run PPSA26578 myself.

  3. MarcZX10 commented on Oct 22, 2026

    @MarcZX10
    Contributor

    The latest update confirms that #8137 is merged and its two exports are present on current main; this closes the reported missing-symbol cause, though the title has not yet been rerun. Could you retry PPSA26578 on a fresh build from current main and share whether it passes the former backtrace stop or reaches a new failure? That result will tell us whether this issue can be closed or needs a separate report. AI-assisted review.

  4. MarcZX10 commented on Oct 23, 2026

    @MarcZX10
    Contributor

    Current upstream main (b9a8704095) now contains implementations of both sceLibcBacktraceGetBufferSize and sceLibcBacktraceSelf for Windows in core/libs/prx/libc/src/Backtrace.cpp; unsupported hosts still throw. This appears to address the missing-export crash reported here at the code level. I have not run PPSA26578, so this is not confirmation that the title now boots. Could you retest on current main and update the report with the result?

  5. EagleBearerYS commented on Oct 25, 2026

    @EagleBearerYS
    Contributor

    @MurphylawdenINC we ran MOUSE: P.I. For Hire (PPSA26578) on current main (2742004), with our own dump, on Windows 11 with an RTX 3070 Laptop GPU. It gets past both stops: the four libc imports from #7624, and the sceLibcBacktraceGetBufferSize abort here. It reaches the first level and plays there with no fatal error.

    • The privacy notice and the EULA need left-stick scrolling to their end. Until then, Cross on "NEXT" does nothing.
    • After them: language and subtitles, screen corners, brightness, the accessibility settings, the title screen, the opening credits and a cutscene. Then comes the first mission ("BIG MOUSE, LITTLE HOPE"), starting in the hotel lobby.
    • Playing: the character walks and turns on the sticks, the compass heading and the view follow, and the HUD stays up. We walked for about 12 minutes, with no stop, device loss or new error in the log. The game was still running when the run ended at 22 minutes.
    • Speed: about 21 fps while walking (15–29), 10–13 standing in the lobby, and about 50 during the credits. The cutscene dropped to about 4. GPU clocks weren't held, so these are rough.
    • Memory: private memory steps up to about 10 GB as the level loads, then stays near that, growing about 13 MB a minute.

    The [exception] guest context signal=0x1e lines: 0x1e is signal 30, SIGUSR1 in FreeBSD's numbering. Each line is that signal being delivered to a guest thread that was waiting. The host records the context at the same place for every delivery, which is why the RIP repeats. They aren't errors: the walking run logged about 2,400 of them in 22 minutes and kept going.

    If a build of current main gets you to the same place, this issue and #7624 can be closed. A new stop is worth its own issue, with its log line.

  6. oneandonlydean commented on Oct 25, 2026

    @oneandonlydean
    Collaborator

    Fixed by #8137 (c8301bb), which implements sceLibcBacktraceSelf and sceLibcBacktraceGetBufferSize on Windows; confirmed above on current main with this title reaching gameplay. Closing. A different stop would be a new issue.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions